A new approach to solid-state nucleation in kinetically-constrained systems

Nucleation is the first step of the phase transformations that we use to control the microstructures of engineering materials. The starting point for questions of nucleation is usually Classical Nucleation Theory (CNT) but for solid-state nucleation at low temperatures where atomic mobility is limited, such as in engineering alloys, CNT has not been very successful is quantitatively predicting nucleation. A strong assumption of CNT is that all thermally-induced stochastic fluctuations, no matter how far their compositions lie from the bulk alloy composition, are possible and that they become nuclei when a critical size determined from thermodynamics is reached. Here we present a new and complementary model for solid-state nucleation. We consider the other extreme where atomic mobility is limited and thermally-induced stochastic clusters cannot form in the time scale relevant for a nucleation event. Instead, we consider the geometric clusters that are a statistical feature of any solution as the origin of the nuclei and present a simple model for the number of nuclei and their rate of ‘activation’. This new ‘geometric cluster’ model is shown to be able to successfully predict the competition in phase nucleation during the crystallization of a series of Al-Ni-Y metallic glass, predict the solvent trapping that is increasingly seen in solid-state nucleation and predict the peak number density of precipitates observed in Cu-Co and Fe-Cu alloys.